Begg 7th edn Ch 04

Perfect competition and pure monopoly are useful benchmarks of the extremes of marketstructure. Most markets are between the

extremes. What determines the structure of a particular market? Why are there 10 000 florists but only a few chemical producers? How does thestructure of an industry affect the behaviour of itsconstituent firms?

A perfectly competitive firm faces a horizontaldemand curve at the market price. It is a price-taker.Any other type of firm faces a downward slopingdemand curve for its product and is imperfectly competitive.

For a pure monopoly, the demand curve for the firmand the industry coincide. We now distinguish twointermediate cases of an imperfectly competitivemarket structure.

An imperfectly competitive firm faces a down-sloping demand curve. Its output price reflectsthe quantity of goods it makes and sells.

The car industry is an oligopoly. The price ofRover cars depends not only on its own outputand sales, but also the output of Ford and Toyota.The corner grocer’s shop is a monopolisticcompetitor. Its output is a subtle package ofphysical goods, personal service, and conveniencefor local customers. It can charge a slightly higherprice than an out-of-town supermarket. But, if itsprices are too high, even local shoppers travel tothe supermarket.

As with most definitions, the lines between differentmarket structures can get blurred. One reason isambiguity about the relevant definition of the market. Is Eurostar a monopoly in cross-channeltrains or an oligopolist in cross-channel travel?

An oligopoly is an industry with few producers,each recognizing their interdependence. Anindustry with monopolistic competition hasmany sellers of products that are close substitutesfor one another. Each firm has only a limited ability to affect its output price.

CHAPTER

9Market structure andimperfect competition

LEARNING OUTCOMES

By the end of this chapter, you should understand:

● How cost and demand affect market structure

● How globalization changes domestic market structure

● Monopolistic competition

● Oligopoly and interdependence

● The kinked demand curve model

● Game theory and strategic behaviour

● Commitment and credibility

● Reaction functions and Nash equilibrium

● Cournot and Bertrand competition

● Stackelberg leadership

● Contestable markets

● Innocent and strategic entry barriers

Begg 7th edn Ch 09 4/18/02 12:29 pm Page 119

Similarly, when a country trades in a competitiveworld market, even the sole domestic producer mayhave little influence on market price. We can neverfully remove these ambiguities, but Table 9–1 showssome things to bear in mind as we proceed throughthis chapter. The table includes the ease with whichnew firms can enter the industry, which affects theability of existing firms to maintain high prices andsupernormal profits in the long run.

9–1 Why market structures differSome industries are legal monopolies, the solelicensed producers. Patent laws may confer temporarymonopoly on producers of a new process. Ownershipof a raw material may confer monopoly status on asingle firm. We now develop a general theory of howdemand and cost interact to determine the likelystructure of each industry.

The car industry is not an oligopoly one day but perfectly competitive the next. Long-run influencesdetermine market structures. Eventually, one firm

can hire another’s workers and learn its technicalsecrets.

Figure 9–1 shows the demand curve DD for the output of an industry in the long run. Suppose allfirms and potential entrants face the average costcurve LAC1. At the price P1, free entry and exit meansthat each firm produces q1. With the demand curveDD, industry output is Q1. The number of firms in the industry is N1 (� Q1/q1). If q1, the minimumaverage cost output on LAC1 is small relative to DD, N1 will be large. Each firm has a tiny effect onindustry supply and market price. We have found aperfectly competitive industry.

Next, suppose that each firm has the cost curve LAC3.Scale economies are vast relative to the market size.At the lowest point on LAC3, output is big relative tothe demand curve DD. Suppose initially two firmseach make q2. Industry output is Q2. The marketclears at P2 and both firms break even. If one firmexpands a bit, its average costs fall. Its higher outputalso bids the price down. With lower average costs,

120 PART 2 Positive microeconomics

Competition Number of firms Ability to affect price Entry barriers Example

Perfect Lots Nil None Fruit stallImperfect: Monopolistic Many Little Small Corner shop

Oligopoly Few Medium Bigger CarsMonopoly One Large Huge Post Office

Table 9–1 Market structure

DD is the industry demand curve. In a competitive industry, minimum efficient scale occurs at an output level q1, when firms have average cost curves LAC1.The industry can support a very large number of firms whose total output is Q1 at the price P1. When LAC3 describes average costs, the industry will be a natural monopoly. When a single firm produces the entire industry output, no other firm can break into the market and make a profit. For intermediatepositions such as LAC2 the industry can support a few firms in the long run, and no single firm can profitably meet the entire demand. The industry will be anoligopoly.

Figure 9–1 Demand, costs, and market structure

Output q of a firm and output Q of the industry

0

Pric

e, a

vera

ge c

osts

P3

P2

P1

q1 q2 q4 Q3 Q2 Q1

LAC3

LAC2

LAC1

D

D


that firm survives but the other firm loses money.The firm that expands undercuts its competitor, anddrives it out of business.

This industry is a natural monopoly. Suppose Q3 isthe output at which its marginal cost and marginalrevenue coincide. The price is P3 and the naturalmonopoly makes supernormal profits. There is noroom in the industry for other firms with access to the same LAC3 curve. A new entrant needs a bigoutput to get average costs down. Extra output onthis scale so depresses the price that both firms makelosses. The potential entrant cannot break in.

Finally, we show the LAC2 curve with moreeconomies of scale than a competitive industry butless than a natural monopoly. This industry supportsat least two firms enjoying scale economies near thebottom of their LAC2 curves. It is an oligopoly.Attempts to expand either firm’s output beyond q4

quickly meet decreasing returns to scale and preventa firm driving competitors out of business.

The crucial determinant of market structure is minimum efficient scale relative to the size of the total market as shown by the demand curve.Table 9–2 summarizes our analysis of the interactionof market size and minimum efficient scale. Whenthe demand curve shifts to the left, an industry previously containing many firms may have room for only a few. Similarly, a rise in fixed costs, raisingthe minimum efficient scale, reduces the number of firms. In the 1950s there were many European aircraft makers. Today, the research and developmentcosts of a major commercial airliner are huge. Apartfrom the co-operative European venture AirbusIndustrie, only the American giant Boeing-McDonnell-Douglas survives.

Monopolistic competition lies between oligopolyand perfect competition. Monopolistic competitorssupply different versions of the same product, suchas the particular location of a newsagent.

Minimum efficient scale is the lowest output atwhich a firm’s LAC curve stops falling.

A natural monopoly enjoys such scaleeconomies that it has no fear of entry by others.

Evidence on market structureThe larger the minimum efficient scale relative to themarket size, the fewer are the number of plants – andprobably the number of firms – in the industry. Whatthe number of plants (NP) operating at minimumefficient scale does a market size allow? Chapter 7discussed estimates of minimum efficient scale indifferent industries. By looking at the total purchasesof a product we can estimate market size. Hence wecan estimate NP for each industry.

Even industries with only a few key players have somesmall firms on the fringe. The total number of firmscan be a misleading indicator of the structure of theindustry. Economists use the N-firm concentration ratioto measure the number of key firms in an industry.

Thus, the 3-firm concentration ratio tells us the market share of the largest three firms. If there arethree key firms, they will supply most of the market.If the industry is perfectly competitive, the largestthree firms will only have a tiny share of industry output and sales.

It would be nice to look at cross-country evidence tosee if market structures always obey our theory. If thisis to be an independent check, we really need nationaldata before globalization and European integrationbecame important. Table 9–3 examines evidence forthe UK, France, and Germany for the mid-1970s.

CR is the 3-firm concentration ratio, the market shareof the top three firms. NP is the number of plants atminimum efficient scale that the market size allows.If our theory of market structure is correct, industrieswith large-scale economies relative to market size,and thus a small number of plants NP, should have alarge concentration ratio CR. Such industries should

The N-firm concentration ratio is the marketshare of the largest N firms in the industry.

Chapter 9 Market structure and imperfect competition 121

Minimum efficient scale relative to market size

Tiny Intermediate LargePerfect competition Oligopoly Natural monopoly

Table 9–2 Demand, cost, and market structure

UK France Germany

Industry CR NP CR NP CR NP

Refrigerators 65 1 100 2 72 3Cigarettes 04 3 100 2 94 3Refineries 79 8 60 7 47 9Brewing 47 11 63 5 17 16Fabrics 28 57 23 57 16 52Shoes 17 165 13 128 20 197

Note: Concentration ratio CR is % market share of 3 largest firms; number of plants NP is market size divided by minimum efficient scale.Sources: F. M. Scherer et al., The Economics of Multiplant Operation, HarvardUniversity Press, 1975; and F. M. Scherer, Industrial Market Structure andEconomic Performance, Rand McNally, 1980.

Table 9–3 Concentration and scale economies


have few key firms. Conversely, where NP is very high,economies of scale are relatively unimportant andthe largest three firms should have a much smallermarket share. CR should be low.

Table 9–3 confirms that this theory of market struc-ture fits these facts. Industries such as refrigeratorand cigarette manufacture had room for few plantsoperating at minimum efficient scale; these indus-tries had high degrees of concentration. The largestthree firms controlled almost the whole market. Scaleeconomies still mattered in industries such as brewing and petroleum refining: the top three firmshad about half the market. Industries such as shoemaking quickly met rising average cost curves;they had room for many factories operating at minimum efficient scale, and thus were much closerto competitive industries. The top three firms inshoemaking had under one-fifth of the market.

Globalization and multinationalsTable 9–3 showed data before the rise of globaliz-ation and multinationals.

Globalization reflects cheaper transport costs, betterinformation technology, and a deliberate policy ofreducing cross-country barriers in order to get efficiency gains from large scale and specialization.

Globalization is the closer integration of marketsacross countries. Multinationals are firms operating in many countries simultaneously.

Multinationals sell in many countries at the sametime. They may, or may not, also produce in manycountries.

Multinationals affect the analysis implied by Figure9–1 and Table 9–3. To what market size should wecompare minimum efficient scale to estimate thenumber of plants that can survive in the long run?Multinationals can produce on a large scale some-where in the world, where production is cheapest,enjoy all the benefits of scale economies, but still sellsmall amounts in many different markets.

This has three effects. First, it reduces entry barriersin a particular country. A foreign multinationalentrant need not achieve a large market share, andtherefore need not bid down the price a lot, to achievescale economies. These now arise because of successin selling globally. Second, small domestic firms, previously sheltered by entry barriers, now facegreater international competition and may not survive. Third, greater competition by low-cost producers leads initially to lower profit margins andlower prices.

However, if there are only a few multinationals, theymay drive the higher cost domestic firms out of business but then collude among themselves to raiseprices again. Some of the debate about globalizationhinges on which of these two outcomes dominates:the initial price fall or a possible subsequent priceincrease. We return shortly to the analysis of collusion. First, we study a simpler case.


Box 9–1 Packaging holidays

In the UK the market for package summer holidays isnow worth £7bn a year as people jet off in search ofsand and sun. Whereas in 1986 the top five chains oftravel agents had a combined market share of 25%,within two years it had soared to 87%. Evidence of hugeeconomies of scale? Not necessarily.

The industry integrated vertically, as travel agents (retailoutlets) combined with tour operators (supplying airline and hotel services). Vertical integration can cutcosts by allowing better coordination between differentstages of the production chain, but it can also enhancemarket power. The two largest tour operators Thomsonand Airtours bought the two largest travel agents (LunnPoly and Going Places). Together, these two firms market share rose to 49%.

The market leaders were accused of unfair practices. In 1996 Lunn Poly refused to display brochures of

First Choice for 4 months until a new agreement oncommissions for travel agents was reached. Small operators had to pay up to 19% commission to Lunn Poly, while Thomson paid only 10%. Thomson and Airtours argued that their size let them keep prices lower, and that smaller competitors couldn’tcompete.

Thomson 28Airtours 21Thomas Cook 19First Choice 17Other 15

Source: The Observer, 19 September 1999.

UK pagkage tours (% market share 1998)


9–2 Monopolistic competitionThe theory of monopolistic competition envisages a large number of quite small firms so that each firmcan neglect the possibility that its own decisions provoke any adjustment in other firms’ behaviours.We also assume free entry and exit from the industryin the long run. In these respects, the industry resembles perfect competition. What distinguishesmonopolistic competition is that each firm faces adownward-sloping demand curve.

Monopolistic competition describes an industry inwhich each firm can influence its market share tosome extent by changing its price relative to its competitors. Its demand curve is not horizontalbecause different firms’ products are only limitedsubstitutes, as in the location of local shops. A lowerprice attracts some customers from another shop,but each shop always has some local customers forwhom convenience is more important than a fewpence on the price of a jar of coffee.

Monopolistically competitive industries exhibit product differentiation. Corner grocers differentiate bylocation, hairdressers by customer loyalty. The special feature of a particular restaurant or hair-dresser lets it charge a slightly different price fromother firms in the industry without losing all its customers.

Monopolistic competition requires not merely product differentiation, but also limited oppor-tunities for economies of scale. Firms are small. Withlots of producers, each can neglect its interdepen-dence with any particular rival. Many examples ofmonopolistic competition are service industrieswhere economies of scale are small.

The industry demand curve shows the total outputdemanded at each price if all firms in the industrycharge that price. The market share of each firmdepends on the price it charges and on the number offirms in the industry. For a given number of firms, ashift in the industry demand curve shifts the demand curve for the output of each firm. For a givenindustry demand curve, more (fewer) firms in theindustry shifts the demand curve of each firm to theleft (right) as its market share falls (rises). But eachfirm faces a downward-sloping demand curve. For agiven industry demand curve, number of firms, andprice charged by all other firms, a particular firm canraise its market share a bit by charging a lower price.

Figure 9–2 shows a firm’s supply decision. Given itsdemand curve DD and marginal revenue curve MR,the firm makes Q0 at a price P0 making short-runprofits Q0 (P0 � AC0). In the long run, these profitsattract new entrants, diluting the market share of

each firm in the industry, shifting their demandcurves to the left. Entry stops when each firm’sdemand curve shifts so far left that price equals average cost and firms just break even. In Figure 9–2this occurs when demand is DD'. The firm makes Q1

at a price P1 in the tangency equilibrium at F.

Note two things about the firm’s long-run equilib-rium at F. First, the firm is not producing at minimumaverage cost. It has excess capacity. It could reduceaverage costs by further expansion. However, its marginal revenue would be so low that this is unprofitable. Second, the firm has some monopolypower because of the special feature of its particularbrand or location. Price exceeds marginal cost.

This explains why firms are usually eager for new customers prepared to buy additional output at theexisting price. We are a race of eager sellers and coybuyers. It is purchasing agents who get Christmaspresents from sales reps, not the other way round. Incontrast, a perfectly competitive firm does not care ifanother buyer shows up at the existing price. With

In monopolistic competition, in the long-runtangency equilibrium each firm’s demand curvejust touches its AC curve at the output level atwhich MC equals MR. Each firm maximizes profits but just breaks even. There is no moreentry or exit.


In the short run the monopolistic competitor faces the demand curve DD and sets MC equal to MR to produce Q0 at a price P0. Profits are Q0 � (P0 � AC0). Profits attract new entrants and shift each firm’sdemand curve to the left. When the demand curve reaches DD' we reachthe long-run tangency equilibrium at F. The firm sets MC equal to MR' toproduce Q1 at which P1 equals AC1. Firms are breaking even and there isno further entry.

Figure 9–2 Equilibrium for a monopolisticcompetitor

Quantity

Pric

e, c

ost

Q1 Q0

P1 = AC1

P0

F

AC0

MC

AC

DD

MRDD'MR'


price equal to marginal cost, the firm is already selling as much as it wants.

9–3 Oligopoly and interdependenceUnder perfect competition or monopolistic com-petition, there are many firms in the industry. Eachfirm can ignore the effect of its own actions on rivalfirms. However, the key to an oligopolistic industry isthe need for each firm to consider how its ownactions affect the decisions of its relatively few competitors. Each firm has to guess how its rivals will react. Before discussing what constitutes a smart guess we introduce the basic tension betweencompetition and collusion when firms know thatthey are interdependent.

Initially, for simplicity, we neglect the possibility ofentry and focus on existing firms.

The profits from collusionAs sole decision-maker in the industry, a monopolistwould choose industry output to maximize totalprofits. Hence, the few producers in an industry can maximize their total profit by setting their totaloutput as if they were a monopolist.

Figure 9–3 shows an industry where each firm, and the whole industry, has constant average andmarginal costs at the level PC. Chapter 8 showed thata competitive industry produces QC at a price PC buta multi-plant monopolist maximizes profits by making QM at a price PM. If the oligopolists collude toproduce QM they act as a collusive monopolist. Having

Collusion is an explicit or implicit agreement toavoid competition.

decided industry output, the firms agree how toshare total output and profits among themselves.

However, it is hard to stop firms cheating on the collective agreement. In Figure 9–3 joint profit ismaximized at a total output QM and price PM. Yeteach firm can expand output at a marginal cost PC.Any firm can expand output, selling at a little belowthe agreed price PM, and make extra profit since itsmarginal revenue exceeds its marginal cost. This firmgains at the expense of its collusive partners. Industryoutput is higher than the best output QM, so totalprofits fall and other firms suffer.

Oligopolists are torn between the desire to collude, tomaximize joint profits, and the desire to compete, toraise market share and profits at the expense of rivals.Yet if all firms compete, joint profits are low and nofirm does very well. Therein lies the dilemma.

CartelsCollusion between firms is easiest if formal agree-ments are legal. Such arrangements, called cartels,were common in the late 19th century, agreeing market shares and prices in many industries. Cartelsare now outlawed in Europe, the United States, andmany other countries. There are big penalties forbeing caught, but informal agreements and secretdeals are sometimes discovered even today.

Cartels across continents are harder to outlaw. Themost famous cartel is OPEC, the Organization ofPetroleum Exporting Countries. Its members meetregularly to set price and output. Initially, OPEC succeeded in organizing quantity reductions to forceup the price of oil. Real OPEC revenues rose 500 percent between 1973 and 1980. Yet many economistspredicted that OPEC, like most cartels, would


Box 9–2 War games

Nintendo, Sony and Microsoft are pitting their videogame consoles against each other, fighting for a a globalindustry now worth £12 billion a year. Sony will spend£500 million in 2002 protecting its huge PlayStationfranchise. Microsoft will spend even more launching itsXbox. Merrill Lynch estimates that Nintendo’s 2001profits will fall by a quarter because of money spentlaunching the GameCube. In the USA alone, Sony hopesto have sold at least 7 million PlayStation 2 consolesbefore its rivals’ consoles come out.

However, the key to success is not hardware but software. Microsoft’s problem is that Sony and Nintendo have a history of popular games, such as theTwisted Metal and Mario series. As a late entrant,

Microsoft needs to overcome these barriers. It is bettingon the Xbox’s online capability, allowing players to compete remotely over the internet. Whereas, withpractice, people can learn to beat the artificial intelligence of a computer, Microsoft hopes that playingunpredictable humans online will be much more interesting.

There is also a bigger picture. The Xbox may beMicrosoft’s Trojan horse. The ultimate fight may not beabout games consoles but set-top boxes. Sony andMicrosoft are battling for control of the entire homeentertainment industry.

Source: Financial Times, 19 May 2001


quickly collapse. Usually, the incentive to cheat is toostrong to resist, and once somebody breaks ranksothers tend to follow.

One reason that OPEC was successful for so long wasthe willingness of Saudi Arabia, the largest oil producer, to restrict its output further when smallermembers insisted on expansion. By 1986 Saudi Arabia was no longer prepared to play by these rules,and refused to prop up the price any longer. The oilprice collapsed from just under $30 to $9 a barrel.During 1987–98, apart from a brief period during theGulf War, oil prices fluctuated between $8 and $20 a barrel. Only after 1998 did OPEC recover the cohesion it displayed during 1973–85.

The kinked demand curveCollusion is much harder if there are many firms inthe industry, if the product is not standardized and ifdemand and cost conditions are changing rapidly. Inthe absence of collusion, each firm’s demand curvedepends on how competitors react. Firms must guesshow their rivals will behave.

Suppose that each firm believes that its own price cutwill be matched by all other firms in the industry, butthat a rise in its own price will not induce a priceresponse from competitors. Figure 9–4 shows thedemand curve DD that each firm then faces. At thecurrent price P0 the firm makes Q0. If competitors donot follow suit, a price rise leads to a large loss ofmarket share to other firms. The firm’s demand curveis elastic above A at prices above the current price P0.However, any price cut is matched by other firms,

and market shares are unchanged. Sales rise onlybecause the whole industry moves down the marketdemand curve as prices fall. The demand curve DD is much less elastic for price cuts from the initial price P0.

In Figure 9–4 marginal revenue MR is discontinuousat the output Q0. Below Q0 the elastic part of thedemand curve is relevant. At the output Q0 the firmhits the inelastic portion of its kinked demand curve and marginal revenue drops. Q0 is the profit-maximizing output for the firm, given its belief abouthow competitors respond.

Suppose the MC curve of a single firm shifts up ordown by a small amount. Since the MR curve has a discontinuous vertical segment at Q0, it remainsoptimal to make Q0 and charge the price P0. In contrast, a monopolist facing a continuously downward-sloping MR curve would adjust quantityand price when the MC curve shifted. The kinkeddemand curve model may explain the empirical finding that firms do not always adjust prices whencosts change.

It does not explain what determines the initial priceP0. One interpretation is that it is the collusivemonopoly price. Each firm believes that an attemptto undercut its rivals will provoke them to co-operateamong themselves and retaliate in full. However, its


By colluding to restrict industry output Qm, joint profits are maximizedand equal to those which a multi-plant monopolist would obtain. Buteach firm, with a marginal cost of Pc, has an incentive to cheat on the collusive agreement and expand its output.

Figure 9–3 Collusion versus competition

AC = MC

DD

MR

Qm

Pc

Quantity

Pric

e, c

ost,

rev

enue Pm

An oligopolist believes rivals will match price cuts but not price rises. Theoligopolist’s demand curve is kinked at A. Price rises lead to a large loss of market share, but price cuts increase quantity only by increasingindustry sales. Marginal revenue is discontinuous at Q0. The oligopolistproduces Q0, the output at which MC crosses the MR schedule.

Figure 9–4 The kinked demand curve

Quantity

0

Pric

e, m

argi

nal r

even

ue, m

argi

nal c

ost

Q0

P0A

MCMR

DD

M


rivals will be happy for it to charge a higher price andsee it lose market share.

If we interpret P0 as the collusive monopoly price, wecan contrast the effect of a cost change for a singlefirm and a cost change for all firms. The latter shiftsthe marginal cost curve up for the entire industry,raising the collusive monopoly price. Each firm’skinked demand curve shifts up since the monopolyprice P0 has risen. Hence, we can reconcile the stickiness of a firm’s price with respect to changes inits own costs alone, and the speed with which theentire industry marks up prices when all firms’ costsincrease. Examples of the latter are higher taxes onthe industry’s product, or a union wage increaseacross the whole industry.

9–4 Game theory andinterdependent decisions

A good poker player sometimes bluffs. You can winwith a bad hand if your opponents misread it for agood hand. Similarly, by having bluffed in the pastand been caught, you may persuade opponents to beta lot when you have a terrific hand.

Like poker players, oligopolists try to anticipate theirrivals’ moves to determine their own best action. Tostudy interdependent decision-making, we use gametheory.

The players in the game try to maximize their ownpayoffs. In an oligopoly, the firms are the players andtheir payoffs are their profits in the long run. Eachplayer must choose a strategy. Being a pickpocket is astrategy. Lifting a particular wallet is a move.

As usual, we are interested in equilibrium. In mostgames, each player’s best strategy depends on thestrategies chosen by other players. It is silly to be apickpocket when the police have TV cameras, or tochoose four centre backs when the opponents haveno proven goalscorers.

Nobody then wants to change strategy, since otherpeople’s strategies were already figured into assessing

In Nash equilibrium, each player chooses thebest strategy, given the strategies being followed byother players.

A strategy is a game plan describing how a playeracts, or moves, in each possible situation.

A game is a situation in which intelligent decisions are necessarily interdependent.

each player’s best strategy. This definition of equilib-rium, and its application to game theory, wasinvented by a Princeton University mathematicianJohn Nash.1

Dominant strategiesSometimes, but not usually, a player’s best strategy isindependent of those chosen by others. We beginwith an example in which each player has a dominantstrategy.

Figure 9–5 shows a game2 between the only twomembers of a cartel. Each firm can select a high-output or low-output strategy. In each box of Figure 9–5 the coloured number shows firm A’s profits and the black number, firm B’s profits for thatoutput combination.

When both have high output, industry output ishigh, the price is low, and each firm makes a smallprofit of 1. When each has low output, the outcomeis like collusive monopoly. Prices are high and eachfirm does better, making a profit of 2. Each firm doesbest (a profit of 3) when it alone has high output; theother firm’s low output helps hold down industryoutput and keep up the price. In this situation weassume the low-output firm makes a profit of 0.

A dominant strategy is a player’s best strategywhatever the strategies adopted by rivals.


1 Nash, who battled with schizophrenia, won the Nobel Prize inEconomics for his work on game theory. A film about his life, ABeautiful Mind, starring Russell Crowe, was released in 2002.

2 The game, called the Prisoners’ Dilemma, was first used toanalyse the choice facing two people arrested and in differentcells, each of whom could plead guilty or not guilty to the onlycrime that had been committed. Each prisoner would pleadinnocent if only he or she knew the other would plead guilty.

The coloured and black numbers in each box indicate profits to firms Aand B, respectively. Whether B pursues high or low output, A makes moreprofit going high; so does B, whichever A adopts. In equilibrium both gohigh. Yet both would make greater profits if both went low!

Figure 9–5 The Prisoners’ Dilemma Game

Firm Aoutput

Firm B output

High

Low

1

0

1

3

3

2

0

2

High Low


Now we can see how the game will unfold. Considerfirm A’s decision. It first thinks what to do if firm Bhas a high-output strategy. Firm A will thus be in oneof the two left-hand boxes of Figure 9–5. Firm A getsa profit of 1 by choosing high but a profit of 0 bychoosing low. If Firm A thinks firm B will choosehigh output, firm A prefers high output itself. Butfirm A must also think what to do if firm B chooses alow-output strategy. This puts firm A in one of thetwo right-hand boxes. Firm A still prefers high outputfor itself, which yields a profit of 3 whereas low yieldsa profit of only 2. Firm A has a dominant strategy.Whichever strategy B adopts, A does better to choosea high-output strategy.

Firm B also has a dominant strategy to choose highoutput. If firm B anticipated that firm A will go high,facing a choice of the two boxes in the top row, firmB prefers to go high. If B thinks A will go low, B facesa choice from the two boxes in the bottom row of Figure 9–5, but B still wants to go high. Firm B doesbetter to go high whichever strategy A selects. Bothfirm A and firm B have a dominant strategy to gohigh. Equilibrium is the top left-hand box. Each firmgets a profit of 1.

Yet both firms would do better, getting a profit of 2,if they colluded to form a cartel and both producedlow – the bottom right-hand box. But neither can risk going low. Suppose firm A goes low. Firm B, comparing the two boxes in the bottom row, will thengo high, preferring a profit of 3 to a profit of 2. Andfirm A will get screwed, earning a profit of 0 in thatevent. Firm A can figure all this out in advance, whichis why its dominant strategy is to go high.

This shows vividly the tension between collusion andcompetition. In this example, it appears that the output-restricting cartel will never get formed, sinceeach player can already foresee the overwhelmingincentive for the other to cheat on such an arrange-ment. How then can cartels ever be sustained? Onepossibility is that there exist binding commitments.

If both players in Figure 9–5 could simultaneouslysign an enforceable contract to produce low outputthey could achieve the co-operative outcome in thebottom right-hand box, each earning profits of 2.This beats the top left-hand box, which shows theNash equilibrium of the game when collusion cannotbe enforced. Without a binding commitment, neitherplayer can go low because then the other player goeshigh. Binding commitments, by removing this temptation, let both players go low. Both playersgain.

A commitment is an arrangement, entered intovoluntarily, that restricts future actions.

This idea of commitment is important, and we shallencounter it many times. Just think of all the humanactivities that are the subject of legal contracts, a simple commitment simultaneously undertaken bytwo parties or players.

Although this insight is powerful, its application tooligopoly requires care. Cartels within a country areusually illegal, and OPEC is not held together by acontract enforceable in international law! Is there a less formal way in which oligopolists can avoidcheating on the collusive low-output solution to thegame? If the game is played only once, this is difficult.

Repeated games In the real world, the game isrepeated many times: firms choose output levels dayafter day. Suppose two players try to collude on lowoutput: each announces a punishment strategy. If firmA ever cheats on the low-output agreement, firm Bsays that it will subsequently react by raising its output. Firm A makes a similar promise.

Suppose the agreement has been in force for sometime and both firms have stuck to their low-outputdeal. Firm A assumes that firm B will go low as usual.Figure 9–5 shows that firm A makes a temporary gaintoday if it cheats and goes high. Instead of staying in the bottom right-hand box with a profit of 2, it can move to the top right-hand box and make 3.However, from tomorrow onwards, firm B will alsogo high, and firm A can then do no better than continue to go high too, making a profit of 1 for evermore. But if A refuses to cheat today it can continue to stay in the bottom right-hand box andmake 2 forever. In cheating, A swaps a temporary gainfor a permanent reduction in future profits. Thus,punishment strategies can sustain an explicit cartelor implicit collusion even if no formal commitmentexists.

It is all very well to promise punishment if the otherplayer cheats. But this will affect the other player’sbehaviour only if your threat is credible.

In the preceding example, once firm A has cheatedand gone high, it is then in firm B’s interest to gohigh anyway. Hence a threat to go high if A evercheats is a credible threat.

These insights shed light on the actual behaviour ofOPEC in 1986, when Saudi Arabia dramaticallyraised its output, leading to a collapse of oil prices. Inthe 1980s, other members of OPEC had graduallycheated on the low-output agreement, trusting thatSaudi Arabia would still produce low to sustain a

A credible threat is one that, after the fact, is stilloptimal to carry out.



high price and the cartel’s prestige. They hoped Saudithreats to adopt a punishment strategy were emptythreats. They were wrong. Figure 9–5 shows that,once the others went high, Saudi Arabia had to gohigh too.

9–5 Reaction functionsIn the previous example, in a one-off game eachplayer had a dominant strategy, to produce high out-put whatever its rival did. This led to a poor outcomefor both players, because they were not co-operatingdespite being interdependent. When the game isrepeated, commitments and punishment strategieshelp players co-operate to find an outcome that isbetter for both of them.

In punishing a rival, a player’s actions change inresponse to bad behaviour by the rival. Dominantstrategies are rare. More usually, each player’s bestaction depends on the actual or expected actions ofother players. How a player reacts depend on what itassumes about its rivals’ behaviour. For simplicity weanalyse duopoly in which there are only two players.

Cournot behaviourIn 1838 French economist Augustin Cournotanalysed a simple model of duopoly.

Imagine a duopoly in which both firms have the sameconstant marginal costs MC. Figure 9–6 draws thedecision problem for firm A. If firm A assumes thatfirm B produces 0, firm A gets the whole industrydemand curve D0. This shows what output firm A can sell given the prices that it charges. From this,firm A calculates the marginal revenue MR0, and produces Q0 to equate its marginal cost and marginalrevenue.

If instead firm A assumes that firm B makes 3 units,firm A faces a demand curve D3 obtained by shiftingthe market demand D0 to the left by 3 units. Firm Bgets 3 units and the residual demand is available forfirm A. For this demand curve D3, firm A computesthe marginal revenue curve MR3, and chooses outputQ3 to equate marginal cost and marginal revenue.

Similarly, if firm A expects firm B to make 5 units,firm A shifts D0 to the left by 5 units to get D5, andproduces Q5 in order to equate marginal cost and itsmarginal revenue MR5. The larger the output thatfirm 2 is expected to make and sell, the smaller is theoptimal output of firm A. Q5 is smaller than Q3 whichis smaller than Q0.

In the Cournot model, each firm treats the output of the other firm as given.

By repeating this exercise for every possible belief thatfirm A has about the output of firm B, yields the reac-tion function of firm A

In the Cournot model, a rival’s action is its outputchoice. Figure 9–7 shows the two outputs QA and QB.From Figure 9–6, firm A makes less the more it thinksthat firm B will make. In Figure 9–7 firm A’s optimaloutput choice is the reaction function RA. If firm B isexpected to produce 1 unit less, firm A chooses toraise output by less than 1 unit. This ensures totaloutput falls, raising the price. Because this lets firm Aearn more on its previous output units, it is notworth raising its output by as much as it expects theoutput of B to fall. Equivalently, in Figure 9–6 firmA’s demand curve shifts more than its marginal revenue curve, hence its output rise is smaller thanthe conjectured fall in the output of firm B.

In the duopoly, both firms are the same. Hence firmB faces a similar problem. It makes guesses about theoutput of firm A, calculates the residual demandcurve for firm B, and chooses its best output. Figure9–7 shows the reaction function RB for firm B, whichalso makes less the more that it assumes its rival willproduce.

A firm’s reaction function shows how its optimaloutput varies with each possible action by itsrival.


Assuming firm B makes 0, firm A faces the market demand curve D0 andmaximizes profits by producing Q0 to equate marginal cost and marginalrevenue. If firm B is assumed to make 3 units, firm A faces the residualdemand curve D3 lying 3 units left of D0. Firm A then makes Q3. If firm Bis assumed to make 5 units, firm A faces D5 and makes Q5. Optimal output for firm A is lower the higher the output that it assumes firm B willmake.

Figure 9–6 Cournot behaviour

MC

Q5

P

D0

Q3 Q0 Q

D3

MR0MR3MR5

D5


Along each reaction function, each firm makes itsbest response to the assumed output of the otherfirm. Only in equilibrium is it optimal for the otherfirm to actually behave in the way that has beenassumed. In Nash equilibrium, neither firm wishes toalter its behaviour even after its conjecture about theother firm’s output is then confirmed.

Since both firms face the same industry demandcurve, their reaction functions are symmetric if theyalso face the same marginal cost curves in Figure 9–6.The two firms then produce the same output Q* asshown in Figure 9–7. If costs differed, we could stillconstruct (different) reaction functions, and theirintersection would no longer imply equal marketshares

Suppose the marginal cost curve of firm A now shiftsdown in Figure 9–6. At each output assumed for firmB, firm A now makes more. It moves further downany MR schedule before meeting MC. Hence in Figure9–7 the reaction function RA shifts up, showing firmA makes more output QA at any assumed output QB

of its rival. The new intersection of the reaction functions, say at point F, shows what happens to theNash equilibrium in the Cournot model.

Nash equilibrium is where the two reactionfunctions intersect.

It is no surprise that the output of firm A rises. Whydoes the output of firm B fall? With lower marginalcosts, firm A is optimally making more. Unless firm Bcuts its output, the price will fall a lot. Firm B prefersto cut output a little, in order to prop up the price abit, preventing a big revenue loss on its existing units.

As in our discussion of the Prisoners’ Dilemma gamein Section 9–4, the Nash–Cournot equilibrium doesnot maximize the joint payoffs of the two players.They fail to achieve the total output that maximizesjoint profits. By treating the output of the rival asgiven, each firm expands too much. Higher outputbids down prices for everybody. In neglecting the fact that its own expansion hurts its rival, each firm’s output is too high.

Each firm’s behaviour is correct given its assumptionthat its rival’s output is fixed. But expansion by onefirm induces the rival to alter its behaviour. A jointmonopolist would take that into account and makemore total profit.

Bertrand behaviourTo show how the assumption about rivals’ behavioursaffects reaction functions and hence Nash equilib-rium, consider a different model suggested byanother French economist, Joseph Bertrand.

Each firm decides a price (and hence an output),reflecting the price it expects its rival to set. We couldgo through a similar analysis to the Cournot model,find reaction curves showing how the price set by eachfirm depends on the price set by its rival, and hencefind the Nash equilibrium in prices for the Bertrandmodel. Knowing the equilibrium price, we couldwork out equilibrium quantity. If the firms are identical, again they divide the market equally.

However, in the Bertrand model, it is easy to see whatthe Nash equilibrium must be. It is the perfectly competitive outcome! Price equals marginal cost.How do we know?

Suppose firm B sets a price above its marginal cost.Firm A can grab the whole market by setting a price alittle below that of firm B. Since firm B can anticipatethis, it must set a lower price. This argument keepsworking until, in Nash equilibrium, both firms priceat marginal cost and split the market between them.There is then no incentive to alter behaviour.

Comparing Bertrand and CournotUnder Bertrand behaviour, Nash equilibrium entailsprice equal to marginal cost, so industry output is

In the Bertrand model of oligopoly, each firmtreats the prices of rivals as given


R4 is the reaction function of firm A, showing how its optimal outputvaries with the output it assumes firm B will make and sell. Since firms are similar, RB is the similar reaction function for firm B, showing its best output given the assumed output by firm A. With these Cournotassumptions about its rival’s behaviour, point E is the Nash equilibrium.Each firm’s guess about its rival’s behaviour is then correct, and neitherfirm wishes to change its behaviour. If the firms are identical, their reaction functions are asymmetric, and both make the same output Q*.

Figure 9–7 Nash–Cournot equilibrium

Q*

QA

RB

F

E

RA

QB

Q*


high. Under Cournot behaviour, Nash equilibriumentails lower industry output and a higher price.Because marginal and average costs are constant,each firm makes profits since the price is higher. Butthe firms do not co-operate. A joint monopolistwould make more profit by co-ordinating outputdecisions. Industry output would be even lower, andthe price even higher.

Thus, Nash equilibrium depends on the particularassumption each firm makes about its rival’s behaviour. Generally, economists prefer the Cournotmodel. In practice, few oligopolies behave like a perfectly competitive industry, as the Bertrand modelpredicts.

Moreover, since prices can be changed rapidly, treating a rival’s price as fixed does not seem plausible.In contrast, we can interpret the Cournot model assaying that firms first choose output capacity, and thenset price. Since capacity takes time to alter, thismakes more sense.

First mover advantageSo far we assumed that the two duopolists make decisions simultaneously. Suppose one firm canchoose output before the other. Does it help to movefirst?

To anticipate how firm B behaves once the output offirm A is fixed, firm A examines the reaction functionof firm B as derived in Figures 9–6 and 9–7. In settingoutput, firm A then takes account of how its ownoutput decisions affect output by firm B.

Firm A thus has a different reaction function. Figure9–7 showed the Cournot reaction function RA treat-ing QB as chosen independently of QA. Now firm Auses the reaction function RB to deduce that a higheroutput QA induces a lower output QB. Hence, Firm Aexpects its own output expansion to bid the pricedown less than under Cournot behaviour. Its mar-ginal revenue schedule is higher up. Firm A knowsthat firm B will help prop up the price by cutting QB

in response to a rise in QA.

Facing a higher MR schedule as a Stackelberg leaderthan under Cournot behaviour, firm A producesmore than under Cournot behaviour. Firm B makesless because it must react to the fact that a high output QA is already a done deal. Firm A ends up with higher output and profits than under Cournot behaviour, but firm B has lower output

In the Stackelberg model, firm B can observe theoutput already fixed by firm A. In choosing out-put, firm A must thus anticipate the subsequentreaction of firm B.

and lower profit. Firm A has a first-mover advantage.

Moving first acts like a commitment that preventsyour subsequent manipulation by the other player.Once firm A has built a large output capacity, firm Ahas to live with the reality that firm A will make largeoutput. The best response of firm B is then low output. By propping up the output price, this helpsfirm A. Being smart, firm A had already figured allthat out.

In some industries, firms are fairly symmetric andCournot behaviour is a good description of howthese oligopolists behave. Other industries have adominant firm, perhaps because of a technical edgeor privileged location. That firm may be able to act asa Stackelberg leader and anticipate how its smallerrivals will then react.

9–6 Entry and potentialcompetition

So far we have discussed imperfect competitionbetween existing firms. To complete our under-standing of such markets, we must also think about the effect of potential competition from new entrants to the industry on the behaviour ofexisting or incumbent firms. Three cases must be distinguished: where entry is trivially easy, where it is difficult by accident, and where it is difficult bydesign.

Contestable marketsFree entry to, and exit from, the industry is a key feature of perfect competition, a market structure in which each firm is tiny relative to the industry.Suppose, however, that we observe an industry withfew incumbent firms. Before assuming that our previous analysis of oligopoly is needed, we mustthink hard about entry and exit. The industry may bea contestable market.

By free entry, we mean that all firms, including bothincumbents and potential entrants, have access tothe same technology and hence have the same costcurves. By free exit, we mean that there are no sunk orirrecoverable costs: on leaving the industry, a firm canfully recoup its previous investment expenditure,including money spent on building up knowledgeand goodwill.

A contestable market has free entry and free exit.

A first-mover advantage means that the playermoving first achieves higher payoffs than whendecisions are simultaneous.



A contestable market allows hit-and-run entry. If theincumbent firms, however few, do not behave as ifthey were a perfectly competitive industry (p � MC �minimum LAC), an entrant can step in, undercutthem, and make a temporary profit before quittingagain.

As globalization proceeds, we should remember thatforeign suppliers are important potential entrants.This can take two forms. First, if monopoly profitsare too high in the domestic market, competitionfrom imports may augment supply, bidding downprices and profits in the domestic market. In theextreme case, in which imports surge in wheneverdomestic prices rise above the world price, we areback in the competitive world analysed in Chapter 8.

Globalization also raises the likelihood that foreignfirms will set up production facilities in the homemarket, a tangible form of entry. By raising the supply of potential entrants, globalization increasesthe relevance of contestable markets as a descriptionof market structure. Moreover, we normally think ofan entrant as having to start from scratch. When anexisting foreign firm enters the domestic market, itsproduction and marketing expertise may already behighly developed.

Globalization may be a two-edged sword. On the onehand, it raises the size of the relevant market andmakes entry easier. On the other hand, by allowingmultinationals to become vast by operating in many countries simultaneously, globalization mayencourage large firms that then have substantial market power wherever they operate. Coke and Pepsiare slugging it out for global dominance, and VirginCola provides only limited competition, even in theUK.

The theory of contestable markets remains contro-versial. There are many industries in which sunk

costs are hard to recover or where the initial expertisemay take an entrant some time to acquire, placing itat a temporary disadvantage against incumbentfirms. Nor, as we shall shortly see, is it safe to assumethat incumbents will not change their behaviourwhen threatened by entry. But the theory does vividlyillustrate that market structure and incumbentbehaviour cannot be deduced simply by counting thenumber of firms in the industry.

In the previous chapter, we were careful to stress thata monopolist is a sole producer who can completely discount fear of entry. We now refine the classificationof Table 9–1 by discussing entry in more detail.

Innocent entry barriersOur discussion of entry barriers distinguishes thosethat occur anyway and those that are deliberatelyerected by incumbent firms.

The American economist Joe Bain distinguishedthree types of entry barrier: product differentiation,absolute cost advantages, and scale economies. Thefirst of these is not an innocent barrier, as we shallshortly explain. Absolute cost advantages, whereincumbent firms have lower cost curves than thosethat entrants will face, may be innocent. If it takestime to learn the business, incumbents will face lowercosts, at least in the short run; if they are smart, theymay already have located in the most advantageoussite. In contrast, if incumbents have undertakeninvestment or R&D specifically with a view to deterring entrants, this is not an innocent barrier. Wetake up this issue shortly.

Figure 9–1 showed the role of scale economies as aninnocent entry barrier. If minimum efficient scale is

An innocent entry barrier is one not deliberatelyerected by incumbent firms.


Box 9–3 Freezing out new entrants?

Unilever is a major player in many consumer productsfrom toothpaste to soap powder. One of its big winners is Wall’s ice cream, which has two thirds of theUK market and generates profits of £100 million a year; retailers’ markups can also be as high as 55%. In addition to established rivals such as Nestlé(www.nestle.com) and Häagen Dazs, Unilever has facednew challenges from frozen chocolate bars. Mars has18% of the market.

A critical aspect of these ‘bar wars’ is the freezer cabinets in which small shops store ice cream. As the

leading incumbent, Unilever ‘loaned’ cabinets free ofcharge to small retailers. Unilever contended that itshigh market share reflected its marketing expertise (justone Cornetto); Mars argued that Unilever erectedstrategic barriers to entry, particularly effective in small shops with space for only one freezer cabinet, byrequiring that only Unilever products were stocked inthe cabinet they loaned to retailers.

In January 2000 the UK government ordered Unilever tostop freezing out competitors.


large relative to the industry demand curve, anentrant cannot get into the industry withoutconsiderably depressing the market price, and it mayprove simply impossible to break in at a profit.

The greater are such innocent entry barriers, themore appropriate it is to neglect potential com-petition from entrants. The oligopoly game thenreduces to competition between incumbent firmsalong the lines we discussed in the previous section.Where innocent entry barriers are low, one of twothings may happen. Either incumbent firms acceptthis situation, in which case competition from potential entrants will prevent incumbent firms from exercising much market power – the outcomewill be closer to that of perfect competition – or else incumbent firms will try to design some entrybarriers of their own.

9–7 Strategic entry deterrenceA strategy is a game plan when decision-making isinterdependent. The word ‘strategic’ is used in everyday language, but it has a precise meaning ineconomics.

In Figure 9–8 a single incumbent firm plays a gameagainst a potential entrant. The entrant can come inor stay out. If the entrant comes in, the incumbentcan opt for the easy life, accept the new rival, andagree to share the market – or it can fight. Fightingentry means producing at least as much as before,and perhaps considerably more than before, so thatthe industry price collapses. In this price war, some-times called predatory pricing by the incumbent, bothfirms do badly and make losses. The top row of boxesin Figure 9–8 shows the profits to the incumbent (inblack) and the entrant (in colour) in each of the threepossible outcomes.

If the incumbent is unchallenged, it does very well,making profits of 5. The entrant of course makesnothing. If they share the market, both make smallprofits of 1. In a price war, both make losses. Howshould the game go?

Suppose the entrant comes in. Comparing the lefttwo boxes of the top row, the incumbent does betterto cave in than to fight. The entrant can figure thisout. Any threat by the incumbent to resist entry is nota credible threat – when it comes to the crunch, it willbe better to cave in. Much as the incumbent wouldlike the entrant to stay out, in which case the

A strategic move is one that influences the otherperson’s choice, in a manner favourable to one’sself, by affecting the other person’s expectationsof how one’s self will behave.

incumbent would make profits of 5, the equilibriumof the game is that the entrant will come in and theincumbent will not resist. Both make profits of 1, thetop left-hand box.

The incumbent, however, may have got its acttogether before the potential entrant appears on the scene. It may be able to invent a binding pre-commitment, forcing itself to resist entry andthereby scare off a future challenge. The incumbentwould be ecstatic if a Martian appeared and guaranteed to shoot the incumbent’s directors if theyever allowed an entry to be unchallenged. Entrantswould expect a fight, would anticipate a loss of 1, and would stay out, leaving the incumbent with apermanent profit of 5.

In the absence of Martians, the incumbent canachieve the same effect by economic means. Supposethe incumbent invests in expensive spare capacitythat is unused at low output. The incumbent has lowoutput in the absence of entry or if an entrant isaccommodated without a fight. Suppose in these situations the incumbent loses 3 by carrying thisexcess capacity. The second row of boxes in Figure9–8 reduces the incumbent’s profits by 3 in these twooutcomes. In a price war, however, the incumbent’soutput is high and the spare capacity is no longerwasted; hence we do not need to reduce the incumbent’s profit in the middle column of boxes inFigure 9–8. Now consider the game again.


In the absence of deterrence, if the entrant enters, the incumbent doesbetter to accept entry than to fight. The entrant knows this and enters.Equilibrium is the top left-hand box, and both firms make a profit of 1.But if the incumbent pre-commits an expenditure of 3 which is recoupedonly if there is a fight, the incumbent resists entry, the entrant stays out,and equilibrium is the bottom right-hand box. The incumbent does better, making a profit of 2.

Figure 9–8 Strategic entry deterrence

Profits without

Entrant

In

Incumbent

Accept Fight

Stay out

1 1 –1 –1 5 0

Profits with –2 1 –1 –1 2 0


If the entrant comes in, the incumbent loses 2 by caving in but only 1 by fighting. Hence entry is resisted. Foreseeing this, the entrant does not enter,since the entrant loses money in a price war. Hencethe equilibrium of the game is the bottom right-handbox and no entry takes place. Strategic entry deterrence has been successful. It has also been profitable. Even allowing for the cost of 3 of carryingthe spare capacity, the incumbent still makes a profitof 2, which is better than the profit of 1 in the top left-hand box when no deterrence was attempted and theentrant came in.

Does deterrence always work? No. Suppose in Figure9–8 we change the right-hand column. In the toprow, the incumbent gets a profit of 3 if no entryoccurs. Without the pre-commitment, the equilib-rium is the top left-hand box as before. But if the incumbent has to spend 3 on a spare capacity pre-commitment, it now makes a profit of 0 in thebottom right-hand box when entry is deterred. Theentrant is still deterred, but the incumbent wouldhave done better not to invest in spare capacity but tolet the entrant in.

This model suggests that price wars should neverhappen. If the incumbent is really going to fight, then

Strategic entry deterrence is behaviour byincumbent firms to make entry less likely.

the entrant should not have entered. This of courserequires that the entrant knows accurately the profitsof the incumbent in the different boxes and thereforecan correctly predict its behaviour. In the real world,entrants sometimes get it wrong. Moreover, if theentrant has much better financial backing than theincumbent, a price war may be a good investment forthe entrant. The incumbent will exit first, and there-after the entrant will be able to cash up and get itslosses back with interest.

Is spare capacity the only pre-commitment availableto incumbents? Pre-commitments must be irrevers-ible, otherwise they are an empty threat; and theymust increase the chances that the incumbent willfight. Anything with the character of fixed and sunkcosts may work: fixed costs artificially increase scaleeconomies and make the incumbent more keen onhigh output, and sunk costs cannot be reversed.Advertising to invest in goodwill and brand loyalty isa good example. So is product proliferation. If theincumbent has only one brand, an entrant may hope to break in with a different brand. But if theincumbent has a complete range of brands or models, an incumbent will have to compete acrossthe whole product range.

9–8 Summing upFew industries in the real world are like the textbookextremes of perfect competition and pure monopoly.


Box 9–4 Why advertise so much?

Advertising is not always meant to erect entry barriers topotential entrants. Sometimes it really does aim toinform consumers by revealing inside information thatfirms have about the quality of their own goods.

When consumers can tell at a glance the quality of aproduct, even before buying it, there is little gain fromadvertising. Black rotten bananas can’t be advertised asfresh. Information is freely available, and attempts todeceive consumers are detected. However, for mostgoods, consumers can’t detect quality before purchase,and discover it only by using the good for a while.

The firm then has inside information over first-time buyers. A conspicuous (expensive) advertising campaignsignals to potential consumers that the firm believes inits product and expects to make enough repeat sales to recoup the fixed cost of initial advertising. Firmswhose lies are quickly discovered don’t invest much inadvertising.

What about goods like refrigerators, essentially a once-off purchase, usually not replaced for a decade or

more? Consumers would really benefit from truthfuladvertising, but producers of high-quality goods haveno incentive to advertise. It would pay lying advertisersto advertise too (since it is ages till they are found out).A willingness to advertise no longer signals the quality ofthe product. So little advertising occurs.

The table below shows advertising spending as a frac-tion of sales revenue for the four types of good identifiedabove. The theory fits the facts rather well.

Quality Time till Example Advertisingdetected buy again as % ofsales

revenue

Before buy Irrelevant Bananas 0.4Soon after buy Soon Biscuits 3.6Long after buy Irrelevant CD player 1.8

Source: E. Davis, J. Kay, and J. Star, ‘Is Advertising Rational?’, BusinessStrategy Review, Autumn, 1991, Oxford University Press.

Advertising spending as a percentage


Most are imperfectly competitive. This chapter introduced you to types of imperfect competition.Game theory in general, and concepts such as commitment, credibility, and deterrence, allow economists to analyse many of the practical concernsof big business.

What have we learned? First, market structure andthe behaviour of incumbent firms are determinedsimultaneously. Economists used to start with a market structure, determined by the extent of scaleeconomies relative to the industry demand curve,then deduce how the incumbent firms would behave(monopoly, oligopoly, perfect competition), thencheck out these predictions against performanceindicators, such as the extent to which pricesexceeded marginal cost. Now we realize that strategicbehaviour by incumbent firms can affect entry, andhence market structure, except where entry is almosttrivially easy.

Second, and related, we have learned the importanceof potential competition, which may come fromdomestic firms considering entry, or from importsfrom abroad. The number of firms observed in theindustry today conveys little information about theextent of the market power they truly exercise. Ifentry is easy, even a single incumbent or apparentmonopolist may find it unprofitable to depart significantly from perfectly competitive behaviour.

Finally, we have seen how many business practices of the real world – price wars, advertising, brand proliferation, excess capacity or excessive researchand development – can be understood as strategiccompetition in which, to be effective, threats must bemade credible by prior commitments.


SUMMARY

● Imperfect competition exists when individual firmsbelieve they face downward-sloping demand curves.The most important forms are monopolisticcompetition, oligopoly, and pure monopoly.

● Pure monopoly status can be conferred bylegislation, as when an industry is nationalized or atemporary patent is awarded. When minimumefficient scale is very large relative to the industrydemand curve, this innocent entry barrier may besufficiently high to produce a natural monopoly inwhich all threat of entry can be ignored.

● At the opposite extreme, entry and exit may becostless. The market is contestable, and incumbentfirms must mimic perfectly competitive behaviour toavoid being flooded by entrants. With anintermediate size of entry barrier, the industry maybe an oligopoly.

● Monopolistic competitors face free entry and exit tothe industry, but are individually small and makesimilar though not identical products. Each haslimited monopoly power in its special brand. Inlong-run equilibrium, price equals average cost butexceeds marginal revenue and marginal cost at thetangency equilibrium.

● Oligopolists face tension between collusion tomaximize joint profits and competition for a largershare of smaller joint profits. Collusion may beformal, as in a cartel, or informal. Without crediblethreats of punishment by its partners, each firmfaces a temptation to cheat.

● Game theory analyses interdependent decisions inwhich each player chooses a strategy. In thePrisoners’ Dilemma game, each firm has a dominantstrategy. With binding commitments, both playerscould do better by guaranteeing not to cheat on thecollusive solution.

● A reaction function shows one player’s bestresponse to the actions of other players. In the Nashequilibrium reaction functions intersect. No playerthen wishes to change his decision.

● In Cournot behaviour each firm treats the output ofits rival as given. In the Bertrand behaviour eachfirm treats the price of its rival as given. TheNash–Bertrand equilibrium entails pricing atmarginal cost. The Nash–Cournot equilibriumentails lower output, higher prices, and profits.However, firms still fail to maximize joint profitsbecause each neglects the fact that its outputexpansion hurts its rivals.

● A firm with a first-mover advantage acts as aStackelberg leader. By deducing the subsequentreaction of its rival, it produces higher output,knowing the rival will then have to produce loweroutput. Moving first is a useful commitment.

● Innocent entry barriers are made by nature, andarise from scale economies or absolute costadvantages of incumbent firms. Strategic entrybarriers are made in boardrooms and arise fromcredible commitments to resist entry if challenged.Only in certain circumstances is strategic entrydeterrence profitable for incumbents.



Review questions

1 An industry faces the demand curve:

(a) Suppose it is a monopolist whose constant MC� 3: what price and output are chosen? (b) Nowsuppose there are two firms, each with MC � AC �3: what price and output will maximize joint profitsif they collude? (c) Why might each firm be temptedto cheat if it can avoid retaliation by the other?

2 With the above industry demand curve, two firms, Aand Z, begin with half the market each whencharging the monopoly price. Z decides to cheat andbelieves A will stick to its old output level. (a) Showthe demand curve Z believes it faces. (b) What priceand output would Z then choose?

3 Vehicle repairers sometimes suggest that mechanicsshould be licensed so that repairs are done only byqualified people. Some economists argue thatcustomers can always ask whether a mechanic wastrained at a reputable institution without needing tosee any licence. (a) Evaluate the arguments for andagainst licensing car mechanics. (b) Are thearguments the same for licensing doctors?

4 Think of five adverts on television. Is their functionprimarily informative, or the erection of entrybarriers to the industry?

5 A good-natured parent knows that childrensometimes need to be punished, but also knowsthat, when it comes to the crunch, the child will belet off with a warning. Can the parent undertake anypre-commitment to make the threat of punishmentcredible?

6 Common fallacies Why are these statementswrong? (a) Competitive firms should get together torestrict output and drive up the price. (b) Firmswouldn’t advertise unless they expected it toincrease sales.

Q 1 2 3 4 5 6 7 8 9 10P 10 9 8 7 6 5 4 3 2 1


Begg 7th edn Ch 04

Documents